Article pubs.acs.org/EF
Steam Reforming of Raw Bio-oil in a Fluidized Bed Reactor with Prior Separation of Pyrolytic Lignin Aingeru Remiro,* Beatriz Valle, A. T. Aguayo, Javier Bilbao, and Ana G. Gayubo Chemical Engineering Department, University of the Basque Country, P.O. Box 644, 48080, Bilbao, Spain ABSTRACT: The effect that operating conditions (temperature, steam/carbon molar ratio, and space-velocity) have on the steam reforming of raw bio-oil has been studied in a two-step reaction unit. In the first step (operated at 500 °C), a carbonaceous solid (pyrolytic lignin) deposits by repolymerization of certain bio-oil components, and the remaining volatiles are reformed in the second step (fluidized bed reactor) on a Ni/La2O3−αAl2O3 catalyst. Under suitable reforming conditions (700 °C, S/C = 9, space-velocity = 8000 h−1), the yields of H2 and CO were 95% and 6%, respectively. Catalyst deactivation was very low, whereby the H2 yield decreased by only 2% over 100 min of reaction. By using dolomite as adsorbent in the reforming reactor, CO2 was effectively captured, and the raw bio-oil was reformed at 600 °C without adding water (S/C = 1.1), thus avoiding its vaporization cost. The yields of H2 and CO were 80−82% and 1%, respectively, for a space-velocity (GC1HSV) of 7000 h−1 and catalyst/ dolomite ratio of 0.25, although a high yield of CH4 (7%) was obtained due to the cracking capacity of the dolomite. The coke content on the catalyst was high (7.7 wt % in 2 h) because of the limited gasification of coke precursors under the operating conditions (low temperature and low S/C ratio) used in the process with CO2 capture.
1. INTRODUCTION The H2 production from fossil sources by reforming of natural gas and petroleum products is increasing, resulting in higher CO2 emissions.1,2 Nowadays, greater H2 production is necessary to satisfy its growing demand, as a fuel and petrochemical raw material. Consequently, new routes from renewable sources (with low CO2 emissions) are required. For this purpose, the lignocellulosic biomass provides different valorization alternatives.3−9 Among these, the steam reforming of the bio-oil obtained by flash pyrolysis of biomass has good prospects for industrial implementation, because this bio-oil is obtained by simple technologies (easy to install and transport in case of delocalized use), which have reached a high development level.10−13 Moreover, the steam reforming avoids the costly separation of water, which is required in other routes for valorizing bio-oil to produce fuel or feedstock in cracking and hydrocracking refinery units.14−17 The stoichiometry of raw bio-oil reforming reaction is:
cracking: CnHmOk → Cx HyOz
methane reforming: CH4 + H 2O ⇔ CO + 3H 2
(5)
2CO ⇔ CO2 + C
(6)
CH4 ⇔ C + H 2
(7)
HxC ⇔ coke precursors (olefins + aromatics) x → C + H2 2
(8)
Knowledge of the fundamentals of bio-oil reforming (such as the behavior of different catalysts based on Ni and noble metals) has been gained in the last two decades, thanks to the reforming studies of model oxygenates (acetic acid, phenol, acetol, etc.)18−25 and bio-oil aqueous fraction,26−32 composed of light oxygenates that are soluble in water.33 However, there are scarce studies on the reforming of raw bio-oil (without prior treatment) because of the repolymerization of certain bio-oil compounds (derived from biomass lignin) that leads to the formation of carbonaceous solid (pyrolytic lignin). This deposition increases the catalyst deactivation, and it can even block the reactor. The previous thermal treatment of the raw bio-oil makes pyrolytic lignin to be deposited isolately. However, this treatment causes a loss of hydrogen content liable to valorization by reforming.34
(1)
This reaction is followed by the water−gas shift reaction (WGS): (2)
Thus, the overall steam reforming can be represented as follows: ⎛ ⎞ m CnHmOk + (2n − k)H 2O → nCO2 + ⎜2n + − k ⎟H 2 ⎝ ⎠ 2
Received: September 13, 2013 Revised: November 11, 2013 Published: November 11, 2013
(3)
Additionally, there are secondary reactions: © 2013 American Chemical Society
(4)
The Boudouard reaction and cracking reactions of CH4 and other hydrocarbons should also be considered because of their role in coke formation.
⎛ ⎞ m CnHmOk + (n − k)H 2O → nCO + ⎜n + − k ⎟H 2 ⎝ ⎠ 2
CO + H 2O ⇔ CO2 + H 2
+ gas (H 2 , CO, CO2 , CH4 , HxC) + coke
7549
dx.doi.org/10.1021/ef401835s | Energy Fuels 2013, 27, 7549−7559
Energy & Fuels
Article
To prevent the pyrolytic lignin deposition, caused by raw bio-oil vaporization, different strategies have been used at laboratory scale: (i) the cofeeding of methanol and bio-oil;35 (ii) the operation in reforming−regeneration cycles (by solid deposits combustion);15,36 and (iii) the operation with two reactors in line, with a controlled deposition in the first step, so that the catalyst suffers lower deactivation in the subsequent step.37−41 In this Article, the latter strategy has been applied by using a first step for pyrolytic lignin repolymerization and a subsequent fluidized bed catalytic reactor for reforming the volatile stream coming from the thermal step. The efficiency of this two-step system for reforming the bio-oil aqueous fraction on a Ni/La2O3−αAl2O3 catalyst was previously proven,42 and the suitable operating conditions for attaining full bio-oil conversion with maximum H2 yield were also determined.43 Moreover, the use of a fluidized bed reactor allows incorporating an adsorbent (dolomite), and thereby the reforming with in situ CO2 capture through CaO carbonation reaction: CaO + CO2 ⇔ CaCO3
Table 1. Composition (wt %) of the Raw Bio-oil and Treated Bio-oil after Thermal Treatment at 500 °C compound/group
raw bio-oil
treated bio-oil
acetic acid acetone 1-hydroxy-2-propanone hydroxyacetaldehyde methanol levoglucosan other ketones other acids other alcohols other aldehydes esters ethers phenols others nonidentified
12.8 5.5 16.3 8.5 1.3 11 3.8 4.0 2.4 6.5 5.1 1.4 16.6 2.3 2.6
15.7 7.2 13.6 5.6 0.3 19.7 3.5 4.8 1.3 5.8 1.9 2.3 14.7 1.9 1.7
2.2. Catalyst and Dolomite. The Ni/La2O3−αAl2O3 catalyst was prepared with 10 wt % of Ni and 9 wt % of La2O3 by using the method described previously by Valle et al.,42 which was established by Alberton et al.49 The La2O3−αAl2O3 support was obtained by impregnation of α-Al2O3, under vacuum at 70 °C, with an aqueous solution of La(NO3)3·6H2O (Alfa Aesar, 99%), followed by drying at 100 °C for 24 h and calcination at 900 °C for 3 h. Subsequently, an impregnation with Ni(NO3)2·6H2O and drying at 110 °C for 24 h were carried out, and the final catalyst was calcined at 700 °C for 3 h. The catalyst was reduced at 700 °C for 2 h, by using a H2−He flow (5 vol % of H2), prior to each reforming reaction. Subsequently, a He flow was established to remove the H2 that might have been retained. The catalyst properties were previously described in detail by Valle et al.42 The physical properties, such as BET surface area (34.8 m2 g−1) and pore volume (0.171 cm3 g−1), were evaluated from the N2 adsorption−desorption isotherms by using a Micromeritics ASAP 2010C analyzer. This device was also used for hydrogen chemisorption measurements for quantifying Ni dispersion and metallic surface, with the resulting values of 5.1% and 3.7 m2 g−1, respectively. The TPO (temperature programmed oxidation) analyses of the coke deposited on the deactivated catalyst were conducted by combustion with air in a Setaram TG-DSC-11 Calorimeter coupled to a mass spectrometer Thermostar Balzers Instrument for monitoring the signals corresponding to mass 18 (H2O) and 44 (CO2). For severely deactivated catalyst samples (coke contents >5 wt %), the coke was analyzed by thermogravimetry (combustion with air) in a TA Instruments Q5000 IR thermobalance. The dolomite used in this Article was provided by Calcinor S.A. (Tolosa, Spain) with particle size 0.8) is obtained with the fresh catalyst in the whole temperature range, with values being higher as the temperature is increased (Figure 2a). The 7552
dx.doi.org/10.1021/ef401835s | Energy Fuels 2013, 27, 7549−7559
Energy & Fuels
Article
of the phenolic derivatives (coming from lignin pyrolysis) of lower molecular weight; and (ii) condensation (activated by the dehydrogenating capacity of the catalyst) of hydrocarbons formed by cracking of bio-oil oxygenates (eq 8). The olefins and aromatics capacity for forming coke is well-known, which is favored with increasing temperature.55 The Boudouard reaction (eq 6) and the CH4 cracking (eq 7) also presumably contribute to the coke formation at high temperature. Coke formation reactions are favored by increasing temperature, but gasification of the coke precursors adsorbed on the catalyst is further enhanced in the range studied. Thereby, coke formation at 700 °C is significantly lower than that at 600 °C (Figure 3). Coke formation is negligible at 800 °C, and, at this temperature, the Ni sintering causes the catalyst deactivation. Accordingly, the catalyst used at 600 and at 700 °C recovers its metallic properties upon coke combustion with air at 600 °C, whereas the catalyst used at 800 °C shows a higher average Ni particle size (12.1 nm) than the fresh catalyst (7.9 nm). 3.2.2. Steam/Carbon Molar Ratio (S/C). The effect of S/C molar ratio was studied by feeding 0.1 cm3 min−1 of raw bio-oil and varying the water flow-rate between 0 and 0.3 cm3 min−1. Figure 5 shows the effect, at 700 °C, on bio-oil conversion, H2 yield (Figure 5a), and the yields of CO and CO2 (Figure 5b). The results remain constant with time on stream as they correspond to a low value of space-velocity (GC1HSV = 32 500 h−1) for which deactivation is negligible (with a coke content
